Principles of Safety-Critical Autonomy

Catalog description

Machine learning has led to tremendous progress in domains such as computer vision, speech recognition, and natural language processing. Fueled by these advances, machine-learning approaches are now being explored to develop intelligent physical systems that can operate reliably in unpredictable environments. These include not only robotic systems such as autonomous cars and drones but also large-scale transportation and energy systems. However, learning techniques widely used today are extremely data-hungry and lack the necessary mathematical framework to provide guarantees of correctness, causing safety concerns as data-driven physical systems are integrated into our society. This course covers the mathematical foundations of dynamical system safety analysis and modern algorithmic approaches for autonomous decision-making in safety-critical contexts. The focus is on designing safe controllers in the presence of system and environment uncertainty. The course will start with an overview of background material from relevant subfields: control theory and robotics. This will be followed by advanced techniques (reachability analysis, Lyapunov and barrier functions, etc.) in this area. The course will conclude with an overview of recent work in ensuring and updating safety guarantees while learning. Project work as part of the course will provide a flavor of research in this new emerging area.